AI safety company and creator of the Claude model family. Anthropic focuses on interpretability research and building AI systems that are safe, honest, and helpful.